.

• Timely diagnosis of infant HIV infection is essential for

antiretroviral therapy initiation, yet about 50% of mother-

infant pairs are lost to follow-up during the postpartum

period.

• We aimed to evaluate real-world effectiveness of an

efficacious two-way theory-based text messaging system

(TextIT) in western Kenya.

• In a pragmatic, cluster randomized, stepped-wedge we

randomly allocated 10 clinics to immediate, and 10 to

delayed implementation

• We used modified Poisson regression with robust

variance estimation to estimate the relative risk and 95%

confidence intervals (CI)

• Generalized estimating equations were applied on

individual-level data to account for clustering by site

Methods

Background

Text messaging for retention in PMTCT: a stepped-wedge cluster-randomized trialThomas A. Odeny1, James P. Hughes2, Elizabeth A. Bukusi1, Eliud Akama1,2, Elvin Geng3, King Holmes2, R. Scott McClelland2

1Kenya Medical Research Institute, Kenya, 2University of Washington, USA, 3University of California San Francisco, USA

Study profile

136 eligible health facilities

116 excluded (bottom 116 facilities by

patient volume) 20 randomized

(top 20 facilities by patient volume)

Phase 1 intervention (N=10 health facilities)

599 Were enrolled

927 Women were screened (1,267 screening visits)

Phase 1 control (N=10 health facilities)

751 Were enrolled

1,411 Women were screened (1,978 screening visits)

569 Were included in the intention-to-treat analysis

713 Were included in the intention-to-treat analysis

Phase 2 intervention (N=20 health facilities)

1,165 Were enrolled

2,343 Women were screened (3,159 screening visits)

1,044 Were included in the intention-to-treat analysis

38 Were excluded 1 Woman died

20 Women had stillbirths 3 Women transferred out

14 Infants died within 8 weeks

121 Were excluded 7 Women died

45 Women had stillbirths 47 Women transferred out 16 Infants died within 8 weeks

6 Missing files

328 Were not eligible 317 Had gestation <28

weeks 11 Declined participation

660 Were not eligible 660 Had gestation <28

weeks

1,178 Were not eligible 1,152 Had gestation <28 weeks

25 Declined participation 1 Shared a phone but had

not disclosed

30 Were excluded 4 Women died 8 Women had stillbirths 1 Woman transferred out

17 Infants died within 8 weeks

Cluster-randomized stepped-wedge trial to show real-world effectiveness of TextIT; Implement with fidelity; SOPs, training guides, checklists;

Delivery(T3)

Theory-based SMS developed from formative qualitative studies (focus groups, motivational interviews)

Discovery(T0 – T1)

RCT to demonstrate efficacy of theory-based SMS (TextIT); Parallel-cohort RCT to assess potential for facility-level dissemination

Development(T2)

Population Health (T4)

Public health impact evaluation of TextIT

Mapping the TextIT strategy onto stages of translational epidemiology

14 text messages tailored for specific points during pregnancy and the postpartum period

• A greater proportion of infants in the intervention group

received HIV testing compared with the control group, but

the difference was small, and not statistically significant

• There was a non-significant increase in maternal postpartum

retention in the intervention periods

• Despite the lack of a significant effect of the intervention, key

lessons emerged, both for strengthening PMTCT and for

implementation research in general

Conclusion

Effect of SMS on infant HIV testing and maternal postpartum retention

* adjusted for intervention time period and randomization stratum; control group as reference; RR=relative risk

Outcome SMS ControlUnadjusted

p-value

Adjusted *p-

valueRR (95% CI)

RR (95% CI)

Infant HIV testing

1,466/1,613 (90.9%)

609/713(85.4%)

1.07 (1.02-1.11) 0.002 1.03

(0.97-1.10) 0.3

Maternal postpartum retention

1,548/1,725 (89.7%)

571/747(76.4%)

1.18 (1.03-1.34) 0.01 1.12

(0.97-1.30) 0.1

Distribution of facilities according to randomization periodBlue=immediate intervention; Red=delayed intervention

Kenya

SMS SoftwareCustom robotic SMS system

Pre-programmed content

Preferred time, language

“Call back” nurse alert

Hi [name]! Congratulations for visiting clinic this week! Please call or flash 0788100133 if you have questions about your pregnancy. We're here to help you!

Hi [name], We wish you a healthy pregnancy and safe delivery! If you would like to plan your delivery, call or flash or send please call me to 0788100133

Dear [name], congratulations on the birth of baby [babyname]! We treasure you both! If you have questions about baby’s health please call or flash 0788100133

Hi [name]! This week, please bring baby [babyname] to clinic for important immunizations to prevent childhood diseases and make sure [babyname] grows up healthy and strong.

Maternal baseline demographic and clinical characteristics.

Results Infant characteristics at birth

* IQR=inter-quartile range

SMS (N=1,684) N (%)

Control (N=695) N (%)

Gestational age at delivery (weeks), median (IQR) 39 (37–41) 39 (36–41)

Live births (vs. stillbirths) 1,629 (96.7) 675 (97.1)

Female 842 (47.7) 328 (43.7)

Birth weight (kg), median (IQR) 3.2 (3.0–3.5) 3.2 (3.0–3.5)

Delivery at health facility (vs. home) 1508 (85.5) 621 (82.7)

Exclusive breastfeeding (vs. other) 1,577 (89.4) 653 (87)

CROI 2018Poster #818